Idea Analyzer

nix-shell -p python3Packages.youtube-transcript-api --run "python3 -c \"
from youtube_transcript_api import YouTubeTranscriptApi
ytt = YouTubeTranscriptApi()
transcript = ytt.fetch('<VIDEO_ID>')
for entry in transcript:
    print(entry.text)
\""

Filter out nix store download noise (lines starting with these, copying, /nix) from stdout before processing.

Fallback chain:

1. nix-shell + youtube-transcript-api (primary — works for any public video, no API key needed)
2. If nix is unavailable, try: yt-dlp --write-auto-sub --skip-download --sub-lang en -o "/tmp/%(id)s" <URL> then read the generated .vtt file
3. If both fail, ask the user to paste the transcript text directly

Never attempt to analyze placeholder/error text. If extraction fails, stop and ask for content.

Extraction Failure Handling

If content extraction fails for any reason (paywall, 403, empty content, broken URL, missing file):

• Inform the user what failed and why
• Ask them to paste the content directly
• Do NOT attempt to analyze error messages or partial/placeholder content

Flags

• --deep — full structured report with competitive landscape, Danish market lens, implementation sketch
• --save — persist analysis as Obsidian note in business/idea-analysis/
• --refresh-context — re-read vault business notes and regenerate references/business-context.md

Flags can appear anywhere in $ARGUMENTS. Strip them before processing content.

Analysis Process

Step 1: Extract Content

Based on detected input type, extract the raw content to analyze. Follow the extraction and failure handling rules above.

Step 2: Identify the Core Idea

Before scoring, extract:

• What is the idea? — one sentence
• Who is it for? — target customer
• How does it make money? — revenue model
• Who's presenting it? — source credibility context

Step 3: Load Context & Criteria

Read evaluation criteria from reference files:

• Hype Detection Patterns
• Business Context — covers BOTH segments (AI consulting + SMB product)
• Scoring Rubric

Important: When assessing dotslash.dev fit, evaluate against BOTH business segments:

• Segment A: AI consulting (strategy, implementation, partnership for tech teams)
• Segment B: Productized SMB offering (chatbots + websites for Copenhagen local businesses) An idea can fit one, both, or neither. Cross-segment fit (serves both) scores highest.

Step 4: Output Results

Analyze in English regardless of input language. If input is in Danish, extract and translate key claims before scoring.

Default (quick verdict):

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
📋 IDEA: [Auto-extracted idea title]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

🔍 Idea Viability     [🟢|🟡|🔴] [Verdict] — [One-line rationale]
⚠️ Hype/BS Score      [🟢|🟡|🔴] [Verdict] — [One-line rationale]
🏢 dotslash.dev Fit   [🟢|🟡|🔴] [Verdict] — [One-line rationale]
💰 Effort vs Payoff   [🟢|🟡|🔴] [Verdict] — [One-line rationale]

💬 One-liner: "[Actionable summary — what to do with this]"

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

For dotslash.dev Fit, specify WHICH segment(s) the idea applies to:

• "🟢 Strong fit (Segment B)" or "🟡 Explore further (both segments)" etc.

Traffic light meanings:

• 🟢 = Strong / Low risk / Strong fit / High payoff
• 🟡 = Explore further / Medium risk / Partial fit / Medium payoff
• 🔴 = Skip / High risk / Poor fit / Low payoff

Verdict labels: "Strong fit", "Explore further", "Skip", "Hype trap", "Steal one piece", "Park for later"

With --deep flag:

Produce the full structured report. See Deep Analysis Template for the complete format.

When using --deep, you MUST:

• Use WebSearch to research the competitive landscape — find real companies doing this
• Use WebSearch to check Danish/Nordic market specifics — local players, demand signals
• Ground all sections in evidence, not speculation
• Evaluate fit against BOTH business segments separately

Step 5: Save (if --save)

If --save flag is present:

1. Create the directory if it doesn't exist: mkdir -p /home/peterstorm/dev/notes/remotevault/business/idea-analysis/
2. Create file at /home/peterstorm/dev/notes/remotevault/business/idea-analysis/YYYY-MM-DD-[slugified-idea-title].md
3. Add frontmatter:

   ---
   date: YYYY-MM-DD
   type: idea-analysis
   source: [URL or "pasted" or filename]
   verdict: [quick verdict summary]
   viability: [🟢|🟡|🔴]
   hype-risk: [🟢|🟡|🔴]
   business-fit: [🟢|🟡|🔴]
   segment-fit: [A|B|both|neither]
   tags: [idea-analysis]
   ---
   

4. Write the full analysis as note content
5. Add [[MOC|Business MOC]] and [[dotslash.dev - Unified Business Identity]] backlinks
6. Confirm save location to user

Refresh Context (--refresh-context)

When invoked with --refresh-context:

1. Read vault business strategy files:
   • business/MOC.md
   • business/smart_website_agency/dotslash.dev - Unified Business Identity.md
   • business/smart_website_agency/plans/dotslash.dev - Master Business Plan 2026.md
   • business/smart_website_agency/pricing/Copenhagen Pricing Strategy - Three-Tier Business Model.md
   • business/smart_website_agency/chatbot/Custom Chatbot Architecture.md
   • business/smart_website_agency/chatbot/Client Acquisition Playbook Denmark.md
   • Also Glob for any new files: business/smart_website_agency/**/*.md
2. Read dotslash.dev website content for consulting positioning:
   • Scan page content files in ~/dev/web/dotslash-dev/src/ for services, philosophy, positioning
3. Synthesize into a balanced business context covering BOTH segments equally
4. Write to this skill's references/business-context.md
5. Confirm completion and summarize any changes detected

Critical Analysis Principles

Be genuinely critical. The user wants a bullshit filter, not validation.

• Default stance is skeptical — ideas must prove themselves
• Weight evidence over claims — show me the revenue, not the promise
• Survivorship bias is everywhere — most "I built X" stories omit the 99% who failed
• Course/info-product sellers have misaligned incentives — they profit from you trying, not succeeding
• "Anyone can do this" is always a lie — execution, timing, and luck matter
• High margins ≠ high probability — 90% margin on zero customers is zero
• If the idea requires being early, check if you're actually early or already late

But be fair. Not everything is hype:

• Some ideas are genuinely good even if the source is annoying
• Separate the idea from the messenger
• "Steal one piece" is a valid verdict — bad source can contain one good insight
• Ideas that fit Segment A (consulting) may look very different from Segment B (product) ideas — evaluate each on its own terms